Patterns of MRI atrophy in tau positive and ubiquitin positive
frontotemporal lobar degeneration
E J Kim, G D Rabinovici, W W Seeley, C Halabi, H Shu, M W Weiner, S J DeArmond, J Q Trojanowski,
M L Gorno-Tempini, B L Miller, H J Rosen
............................................................... ............................................................... .....
J Neurol Neurosurg Psychiatry 2007;78:1375–1378. doi: 10.1136/jnnp.2006.114231
We applied optimised voxel based morphometry (VBM) to
brain MRIs from autopsy proven cases of tau positive
frontotemporal lobar degeneration (FTLD-T, n=6), ubiquitin
and TDP-43 positive/tau negative FTLD (FTLD-U, n=8) and
cognitively normal controls (n=61). The analysis revealed that
FTLD-T and FTLD-U both show atrophy in the frontal cortex and
striatum, but striatal atrophy is more severe in FTLD-T. Manual
region of interest tracing of caudate and putamen volumes
confirmed the VBM findings. These anatomical differences may
help distinguish between FTLD spectrum pathological subtypes
One group shows tau immunoreactive inclusions (FTLD-T)
while the other shows ubiquitin and TDP-43 positive/tau
negative pathology (FTLD-U).1 2As molecule specific therapies
are developed for FTLD, it will become important to accurately
predict pathological subtypes. Neuroimaging could serve this
We applied voxel based morphometry (VBM) to the earliest
brain MRIs of pathologically proven cases of FTLD-T and FTLD-U
in order to search for distinct patterns of tissue loss that could
assist in predicting the pathological subtype in vivo.
he pathology of frontotemporal lobar degeneration (FTLD)
is heterogeneous, with two main divisions based on
immunohistochemical staining of intracellular inclusions.
Fourteen patients (six FTLD-T and eight FTLD-U) (table 1) seen
at the UCSF Memory and Aging Center (MAC) between 1998
and 2006 who met pathological criteria for FTLD at autopsy1
and had good quality MRI scans with no strokes were included.
Patients with a pathological diagnosis of dementia lacking
distinctive histology, corticobasal degeneration or progressive
supranuclear palsy were excluded.
Sixty-one cognitively normal controls with no history of
neurological or psychiatric illness were selected from a cohort of
volunteers followed at the MAC. All patients and controls had
undergone at least one clinical evaluation at the MAC,
including a neurological examination, informant interview
and neuropsychological testing, reviewed elsewhere.7Clinical
diagnosis was based on standard research criteria,8while
controls were determined to be ‘‘cognitively normal’’ after a
comprehensive clinical assessment. Clinical diagnosis was
blinded to imaging findings.
Twelve of 14 autopsies were performed at UCSF or the
University of Pennsylvania using a previously described
protocol.9One autopsy was performed at the University of
California, Irvine, and the other at Stanford University. All
reports were reviewed by a neurologist to ensure adherence to a
standard protocol. Six of eight FTLD-U patients were re-stained
for TDP-43 and found to show TDP-43 immunoreactive
inclusions, which have been reported to be absent in neuronal
intermediate filament inclusion disease.2Therefore, a signifi-
cant contribution of patients with neuronal intermediate
filament inclusion disease to the FTLD-U group is unlikely.
Image acquisition and voxel based morphometry
MRI scans and optimised VBM were performed using
previously described protocols10 11and the SPM2 software
package (http://www.fil.ion.ucl.ac.uk/spm). Comparisons were
made using the following contrasts: (1) FTLD-T versus controls,
(2) FTLD-U versus controls; (3) FTLD-T versus FTLD-U; and (4)
FTLD-U versus FTLD-T, with total intracranial volume, age, sex
and Mini-Mental State Examination (MMSE) as covariates. A
conjunction analysis of contrasts (1) and (2) was performed to
identify voxels where both patient groups showed tissue loss
compared with controls. The conjunction results were inclu-
sively masked with each individual contrast to ensure that the
effect was present in each contrast.12
Voxels were considered significant at p,0.05 after family-
wise error correction for multiple comparisons. Results are
displayed on a study specific template brain as t-maps
thresholded at p,0.001 uncorrected for multiple comparisons.
Mean MMSE score in FTLD-T was significantly lower than in
FTLD-U (table 1). FTLD-T patients were also more impaired on
a modified version of the Trails-B task than FTLD-U patients
(p,0.05, uncorrected for multiple comparisons) but there were
no other differences in cognitive performance. The majority of
patients in both groups presented clinically as frontotemporal
dementia, and five patients in the FTLD-U group had comorbid
motor neuron disease.
Compared with controls, patients with both pathological
subtypes demonstrated decreased gray matter volumes in
frontal lobes (predominantly in the ventromedial and striatal
regions) and anterior temporal lobes (fig 1A). After correction
for multiple comparisons, FTLD-T patients had atrophy in the
right inferior frontal gyrus, left head of caudate/anterior
putamen and right putamen, while FTLD-U patients had
greater atrophy than controls in the left inferior frontal gyrus,
left hippocampus, left caudate and left globus pallidus (table 2).
Abbreviations: FTLD, frontotemporal lobar degeneration; FTLD-T, tau
positive frontotemporal lobar degeneration; FTLD-U, ubiquitin and TDP-43
positive/tau negative frontotemporal lobar degeneration; MAC, Memory
and Aging Center; MMSE, Mini Mental State Examination; VBM, voxel
The conjunction analysis revealed significant volume loss in
the left caudate, bilateral inferior frontal gyrii and right
hippocampus (fig 1B, table 2).
FTLD-T patients showed greater atrophy in the striatum
compared with FTLD-U (fig 1C), with voxels in the right
putamen being significant after correction for multiple com-
parisons (table 2). The contrast of FTLD-U versus FTLD-T did
not yield significant results.
FTLD-T showed significant atrophy in the genu and anterior
body of the corpus callosum and the white matter underlying
the anterior cingulate cortex compared with both controls and
FTLD-U (fig 1C). Significant white matter atrophy was found
underlying the left inferior temporal cortex in FTLD-U
compared with controls (data not shown).
The striatal findings in FTLD-T and FTLD-U could have been
confounded by tissue misclassification in periventricular
structures—a potential problem in VBM.13
confirmed the VBM findings using manually traced striatal
volumes. Using Brains 2 software (http://www.psychiatry.
uiowa.edu/mhcrc/IPLpages/IPLhome.htm), a trained investiga-
tor blinded to clinical and pathological information drew
regions of interest around the left and right caudate and
putamen on the same T1 weighted MRIs used for VBM analysis
for the six FTLD-T patients, 7/8 FTLD-U patients and nine 9/61
controls (selected at random). Landmarks used to identify
regions of interest are listed in appendix 1 (appendix 1 can be
viewed on the J Neurol Neurosurg Psychiatry website at http://
www.jnnp.com/supplemental). Volumes were normalised to
total intracranial volume and analysed with ANCOVA, using
diagnosis as the grouping variable, and age, sex and MMSE as
Compared with controls, patients with FTLD-T had signifi-
cantly lower volumes in the bilateral caudate and putamen (left
caudate: p=0.017; right caudate: p=0.011; left putamen:
p=0.027; right putamen: p=0.001), and FTLD-U patients
showed significantly smaller volumes in the bilateral putamen
(left putamen: p=0.027; right putamen: p=0.020). Mean
right caudate volume was smaller in FTLD-T than in FTLD-U
(p=0.007), with a non-significant trend in the right putamen
Parkinsonian motor features
Based on the finding of greater striatal atrophy in FTLD-T, we
retrospectively investigated whether there were differences in
parkinsonian symptoms and signs between groups. A blinded
neurologist rated the presence (1) or absence (0) of nine
parkinsonian features (slowed speech, decreased facial expres-
sion, tremor at rest, action tremor, rigidity, postural instability,
parkinsonian gait, bradykinesia, decreased leg or hand agility)
based on the transcribed descriptions of the clinical history and
Group characteristics in FTLD-T, FTLD-U and NC
FTLD-T (n)FTLD-U (n)NC p Value
Age at MRI (y)
Onset to MRI (y)
1.3 (0.5) (4)
0.8 (0.6) (7)
CDR, Clinical Dementia Rating; FTD, frontotemporal dementia; FTLD, frontotemporal lobar degeneration; FTLD-T, FTLD
with tau positive inclusions; FTLD-U, FTLD with ubiquitin and TDP-43 positive/tau negative inclusions; MMSE, Mini
Mental State Examination; MND, motor neuron disease; n, number of patients whose data were available for analysis;
NC, normal controls; PNFA, progressive non-fluent aphasia; SD, semantic dementia.
*p,0.05 vs controls.
?p,0.05 vs FTLD-U.
`Four of these patients also had motor neuron disease.
Regions of gray matter loss in FTLD-T and FTLD-U compared with controls and compared with each other
Common atrophic areaFTLD-T vs NC FTLD-U vs NC
X, Y, ZT X, Y, ZT X, Y, ZT
L head of caudate/anterior putamen
R inferior frontal gyrus
L middle OF gyrus
L inferior frontal gyrus
26, 9, 104.30*
213, 15, 3
20, 14, 21
37, 24, 212
29, 13, 10 6.83*
4742, 16, 283.81*
42, 16, 28
217, 4, 0
227, 211, 217
223, 25, 216
32, 210, 215
232, 61, 25
240, 11, 27
223, 25, 216
30, 29, 213
230, 61, 26
228, 33, 213
223, 25, 216
30, 29, 213
244, 58, 26
252, 44, 27
BA, Brodmann area; FTLD-T, FTLD with tau positive inclusions; FTLD-U, FTLD with ubiquitin and TDP-43 positive/tau negative inclusions; L, left; NC, normal controls; OF,
orbitofrontal; R, right; T, T score at given voxel.
*p,0.05 after family-wise error correction for multiple comparisons for this contrast.
?FTLD-T versus FTLD-U in p,0.05 after family-wise error correction for multiple comparisons.
1376Kim, Rabinovici, Seeley, et al
neurological examination in each patient. Almost all patients
(5/5 FTLD-T, 7/8 FTLD-U) had at least one parkinsonian feature
at that time of the MRI. Only FTLD-T patients (2/5) had three
parkinsonian features. Mean score in each group was 1.8 (1.1)
in FTLD-T and 1.4 (0.7) in FTLD-U (p=0.419).
We compared gray and white matter tissue content in the two
major pathological variants of FTLD. Compared with healthy
controls, both FTLD-T and FTLD-U had gray matter loss in the
frontal cortices and striatum bilaterally. Striatal atrophy was
more severe in FTLD-T than in FTLD-U. These findings
highlight a stronger association between striatal damage and
tau based pathology as opposed to ubiquitin/TDP-43 based
There is no prior report of striatal atrophy in pathologically
confirmed FTLD. The finding is consistent with the clinical data
in this cohort indicating a slightly higher burden of parkinson-
ism in FTLD-T, and with the literature relating atypical
parkinsonism with tau positive pathology.9 14–16The worse
performance for FTLD-T on our version of the Trails-B task,
which is timed, could also be a sign of motor slowing. As none
of these patients were diagnosed with parkinsonian disorders,
the fact that FTLD-T only showed slightly more parkinsonism is
not surprising. These data suggest that tau pathology has a
special proclivity for the basal ganglia, even in patients who do
not present with parkinsonian features among their chief
The pathogenesis of white matter atrophy in FTLD has not
been fully defined. However, the loss of tissue in the anterior
corpus callosum in FTLD-T is consistent with previous MRI
work showing anterior callosal atrophy in frontotemporal
dementia relative to other neurodegenerative diseases.17Our
data suggest that this may be particularly true with tau
pathology. In contrast with patients with corticobasal degen-
eration (which is associated with tau pathology), FTLD subjects
in this study lacked abundant white matter pathology at
autopsy (data not shown).18Thus the white matter loss may
frontotemporal lobar degeneration (FTLD)
pathological subtypes. T score maps are
displayed on multiple axial sections on a
study specific template. L, left; R, right. (A)
Gray matter loss in FTLD with tau positive
inclusions (FTLD-T, upper row) and FTLD with
ubiquitin and TDP-43 positive/tau negative
inclusions (FTLD-U, bottom row) compared
with normal controls (NC). (B) Common
regions of gray matter loss found in both
groups compared with controls. (C) Regions
of greater gray matter loss (displayed in
pink) and white matter loss (displayed in sky
blue) in FTLD-T compared with FTLD-U. T
score maps are presented on axial (z=21),
coronal (y=13) and sagittal (x=210)
sections of a study specific template.
Patterns of gray matter loss in
MRI atrophy in tau positive and ubiquitin positive FTLD1377
also relate to demyelination or axon loss within tracts. This
possibility suggests that approaches more appropriate for
studying white matter tracts, such as diffusion tensor imaging,
may be valuable for studying FTLD.19
The major limitation of our study and similar case series3–6is
the small number of patients, a frequent problem in imaging
studies of patients with pathologically confirmed diagnoses.
While our findings need to be confirmed in larger series, we
detected significant differences in atrophy patterns between
FTLD pathological subtypes that converge with clinical features,
suggesting that these distinctive patterns are clinically relevant.
Understanding how FTLD-T and FTLD-U target differing brain
regions may shed light on the underlying pathogenesis of these
closely related diseases.
This study was supported by the John Douglas French Alzheimer’s
Foundation, the Larry L Hillblom Foundation, NIA grants P01-
AG1972403, R01-AG022983, K08-AG027086-01, K08-AG020760-01,
ADRC grant P50-AG023501 and GCRC grant M01-RR00079.
Appendix 1 can be viewed on the J Neurol Neurosurg
Psychiatry website at http://www.jnnp.com/supple-
E J Kim, G D Rabinovici, W W Seeley, C Halabi, H Shu, M L Gorno-
Tempini, B L Miller, H J Rosen, Memory and Aging Center, and
Department of Neurology, University of California, San Francisco, San
Francisco, California, USA
M W Weiner, Department of Neurology and Department of Radiology,
University of California, San Francisco, San Francisco, California, USA,
and Magnetic Resonance Imaging Unit, San Francisco Veterans Affairs
Hospital, San Francisco, California, USA
S J DeArmond, Department of Pathology, University of California, San
Francisco, San Francisco, California, USA
J Q Trojanowski, Center for Neurodegenerative Disease Research and
Institute on Aging, Department of Pathology and Laboratory Medicine,
University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania,
Competing interests: None.
Correspondence to: Dr H J Rosen, UCSF Memory and Aging Center, 350
Parnassus Ave, Suite 706, San Francisco, CA 94117, USA; hrosen@
Received 28 December 2006
Revised 14 May 2007
Accepted 14 June 2007
Published Online First 5 July 2007
1 McKhann GM, Albert MS, Grossman M, et al. Clinical and pathological
diagnosis of frontotemporal dementia: report of the Work Group on
Frontotemporal Dementia and Pick’s Disease. Arch Neurol 2001;58:1803–9.
2 Neumann M, Sampathu DM, Kwong LK, et al. Ubiquitinated TDP-43 in
frontotemporal lobar degeneration and amyotrophic lateral sclerosis. Science
3 Whitwell JL, Warren JD, Josephs KA, et al. Voxel-based morphometry in tau-
positive and tau-negative frontotemporal lobar degenerations. Neurodegener Dis
4 WhitwellJL, JosephsKA,RossorMN,etal.Magnetic resonanceimagingsignatures
of tissue pathology in frontotemporal dementia. Arch Neurol 2005;62:1402–8.
5 Whitwell JL, Jack CR Jr, Senjem ML, et al. Patterns of atrophy in pathologically
confirmed FTLD with and without motor neuron degeneration. Neurology
6 Josephs KA, Whitwell JL, Dickson DW, et al. Voxel-based morphometry in
autopsy proven PSP and CBD. Neurobiol Aging (in press).
7 Kramer JH, Jurik J, Sha SJ, et al. Distinctive neuropsychological patterns in
frontotemporal dementia, semantic dementia, and Alzheimer disease. Cogn
Behav Neurol 2003;16:211–18.
8 Neary D, Snowden JS, Gustafson L, et al. Frontotemporal lobar degeneration: a
consensus on clinical diagnostic criteria. Neurology 1998;51:1546–54.
9 Forman MS, Farmer J, Johnson JK, et al. Frontotemporal dementia:
clinicopathological correlations. Ann Neurol 2006;59:952–62.
10 Good CD, Scahill RI, Fox NC, et al. Automatic differentiation of anatomical
patterns in the human brain: validation with studies of degenerative dementias.
11 Good CD, Johnsrude IS, Ashburner J, et al. A voxel-based morphometric study of
ageing in 465 normal adult human brains. Neuroimage 2001;14:21–36.
12 Friston KJ, Penny WD, Glaser DE. Conjunction revisited. Neuroimage
13 Senjem ML, Gunter JL, Shiung MM, et al. Comparison of different
methodological implementations of voxel-based morphometry in
neurodegenerative disease. Neuroimage 2005;26:600–8.
14 Hodges JR, Davies RR, Xuereb JH, et al. Clinicopathological correlates in
frontotemporal dementia. Ann Neurol 2004;56:399–406.
15 Kertesz A, McMonagle P, Blair M, et al. The evolution and pathology of
frontotemporal dementia. Brain 2005;128:1996–2005.
16 Josephs KA, Petersen RC, Knopman DS, et al. Clinicopathologic analysis of
17 Yamauchi H, Fukuyama H, Nagahama Y, et al. Comparison of the pattern of
atrophy of the corpus callosum in frontotemporal dementia, progressive
supranuclear palsy, and Alzheimer’s disease. J Neurol Neurosurg Psychiatry
18 Forman MS, Zhukareva V, Bergeron C, et al. Signature tau neuropathology in
gray and white matter of corticobasal degeneration. Am J Pathol
19 Larsson EM, Englund E, Sjobeck M, et al. MRI with diffusion tensor imaging post-
mortem at 3.0 T in a patient with frontotemporal dementia. Dement Geriatr Cogn
1378Kim, Rabinovici, Seeley, et al